Acquirers lost $6 billion to merchant and transaction-level fraud in the U.S. in 2019, with an additional $443 million lost to false declines. Amidst a challenging year for business, fraud prevention is more pressing than ever.
In the CNP (card not present) environment, merchants and their acquiring banks face increasing risk from transaction-level fraud. CNP fraud in 2014 cost companies $2.8 billion, rising to $5.5 billion by the end of 2018. Acquirers were hit by $6 billion in fraud losses last year and Aite Group estimates CNP fraud will reach $6.4 billion by 2021. However, with the new economic implications of COVID-19, that estimate now seems low.
In these times of social distancing and working at home, the volume of online retail and digital goods purchases rose 67 percent between March 1 and June 30, 2020, according to analysts at NuData, our sister company under the Mastercard umbrella. However, one in every two account creation attempts was flagged as high risk by the NuData platform during the same period.
AI models enable reduced transaction-level fraud and merchant risk
Common fraud detection solutions are built with rules-based software. Manual updates are onerous and, with such a rapidly changing landscape, it’s almost impossible to keep up with the changing face of fraud. There is a better way.
Brighterion’s adaptive AI models reduce the need for rules. Based on multiple data points, such as transaction and user history, current activity, and account events, models can identify patterns and anomalous behavior that humans can’t possibly do. It’s also done in milliseconds, keeping pace with evolving fraud trends and sophistication.
On the ground, that means acquirers can onboard merchants, monitor account activity, and be alerted when unusual behavior or transactions occur, indicating merchant credit risk or fraudulent transactions. It also means acquirers and merchants will be alerted to transaction-level fraud in progress with the assurance that false declines are vastly reduced by the accuracy of AI’s one-to-one data analysis.
“The payments industry is changing, and it’s changing rapidly,” says Ian Belsham, Global Head of Transaction Monitoring at Worldpay, the largest global acquirer.
Real-world results from Worldpay
Worldpay became a customer of Brighterion in 2011, in advance of the London Olympics. This leading acquirer was specifically interested in Brighterion AI’s ability to gather more insightful, holistic intelligence.
Brighterion quickly built the AI model using Worldpay’s historical data, reducing 50,000 rules to 250. The model was able to start identifying anomalies right away, and base predictions on past transactions.
Rather than looking at just transaction data, Brighterion AI analyzed Worldpay merchants’ geographic data, peer analysis, history and time of transactions to determine if an alert was actually warranted. The more data accrued, the more effective Brighterion became.
With Brighterion’s highly dynamic, real-time AI technology, Worldpay increased its fraud detection and prevention by 3.2 times and incurred 20 times less false positives. And as Worldpay set out on a course of massive growth, Brighterion’s proprietary distributed file storage made the platform inherently scalable.
Benefits of AI for acquirers
With results like these, the benefits become apparent very quickly. The reduction in false positives is paired with improved fraud detection. In Worldpay’s case, transaction-level and merchant fraud detection increased by more than three times.
The cost savings to acquirers for manual review is just one aspect. Reducing false declines at the source improves customer and merchant experience. Research by Javelin strategies found that 44 percent of consumers stopped shopping at a retailer after receiving a false decline. That enormous revenue loss has more than bottom-line impacts; it also affects merchant brands amongst consumers.
Access our research and reports on transaction-level fraud and merchant risk, as well as information on how to stop the cycle with adaptive AI.
